Setting parameter NSTEEP to 1 causes LINGO’s nonlinear solver to use steepest-edge variable selection. When LINGO is not in steepest-edge mode, the nonlinear solver will tend to select variables that offer the highest absolute rate of improvement to the objective, regardless of how far other variables may have to move per unit of movement in the newly introduced variable. The problem with this strategy is that other variables may quickly hit a bound, resulting in little gain to the objective. With the steepest-edge option, the nonlinear solver spends a little more time in selecting variables by looking at the rate at which the objective will improve relative to movements in the other nonzero variables. Thus, on average, each iteration will lead to larger gains in the objective. In general, the steepest-edge option will result in fewer iterations. However, each iteration will take longer. LINGO defaults to not using the steepest-edge option.